wemake-ai/mcp

开发工具

本页由 AI 翻译，技术细节请以原文为准。查看原文

6 stars 5 forks 质量 78 趋势 14MIT

Repository Install

wemake-ai/mcp 是从 https://github.com/wemake-ai/mcp. 导入的 MCP 服务器候选

概览

wemake-ai/mcp 是从 https://github.com/wemake-ai/mcp. 导入的 MCP 服务器候选

概述（AI 翻译）

德国市场的AI优先企业解决方案

@florentin-one/mcp 是一个统一的模型上下文协议（MCP）服务器将所有七个 Florentin One 推理工具打包到一个 Cloudflare Worker 中。专为企业环境打造 GDPR合规、德国数据主权和零停机部署。

建筑

一个 @florentin-one/mcp 包通过一个 MCP 端点暴露所有七个推理工具。服务器运行于 Cloudflare Workers 配备持久对象，用于国家管理，仅部署在欧盟数据管辖区内。

MCP规范

实现 MCP 2026-07-28 使用 @modelcontextprotocol/server v2 和 可流式 HTTP 传输 — 完全无状态协议。每个请求都是自包含的;无需会话亲和或粘性路由。

安装

一个包替换了之前七个独立的@florentin-one/mcp-*包。

可用工具

工具	描述
`metacognitiveMonitoring`	系统自我监测：知识边界、置信度校准、偏差检测

本页由 AI 翻译，技术细节请以原文为准。

README

以下为原文 README 节选，完整内容请查看原文。

Florentin One Unified MCP Server

AI-First Enterprise Solutions for the German Market

@florentin-one/mcp is a single, unified Model Context Protocol (MCP) server packaging all seven Florentin One reasoning tools into one Cloudflare Worker. Built for enterprise environments with GDPR compliance, German data sovereignty, and zero-downtime deployment.

Architecture

A single @florentin-one/mcp package exposes all seven reasoning tools through one MCP endpoint. The server runs on Cloudflare Workers with Durable Objects for state management, deployed exclusively within EU data jurisdiction.

┌──────────────────────────────────────────────┐
│  @florentin-one/mcp (single package)         │
│  ┌────────────────────────────────────────┐  │
│  │  MCP Server (Streamable HTTP)          │  │
│  │  metacognitiveMonitoring               │  │
│  │  sequentialthinking                    │  │
│  │  collaborativeReasoning                │  │
│  │  scientificMethod                      │  │
│  │  structuredArgumentation               │  │
│  │  constraintSolver                      │  │
│  │  narrativePlanner                      │  │
│  └────────────────────────────────────────┘  │
│  ┌──────────────┐  ┌──────────────────────┐  │
│  │  Code Mode   │  │  Agent SDK Handler   │  │
│  │  (direct TS) │  │  (createMcpHandler)  │  │
│  └──────────────┘  └──────────────────────┘  │
└──────────────────────────────────────────────┘

MCP Specification

Implements MCP 2026-07-28 using @modelcontextprotocol/server v2 with Streamable HTTP transport — a fully stateless protocol. Each request is self-contained; no session affinity or sticky routing required.

Installation

npm install @florentin-one/mcp

A single package replaces the previous seven individual @florentin-one/mcp-* packages.

Available Tools

Tool	Description
`metacognitiveMonitoring`	Systematic self-monitoring: knowledge boundaries, confidence calibration, bias detection
`sequentialthinking`	Dynamic step-by-step reasoning with revision, branching, and dependency tracking
`collaborativeReasoning`	Multi-persona expert collaboration simulation with structured disagreement resolution
`scientificMethod`	Formal hypothesis testing, variable identification, experiment design, evidence evaluation
`structuredArgumentation`	Dialectical reasoning with thesis-antithesis-synthesis progression
`constraintSolver`	Mathematical constraint satisfaction validation for numeric variables
`narrativePlanner`	Three-act story structure planning with character development

MCP Client Configuration

Cursor

Add to ~/.cursor/mcp.json:

{
  "mcpServers": {
    "Florentin One MCP": {
      "command": "npx",
      "args": ["@florentin-one/mcp@latest"]
    }
  }
}

Claude Desktop

Add to Claude Desktop configuration:

{
  "mcpServers": {

在 GitHub 查看完整 README

安装

npx @florentin-one/mcp@latest

新手？先看 mcpskill.org 教程 →

配置

{
  "mcpServers": {
    "Florentin One MCP": {
      "command": "npx",
      "args": ["@florentin-one/mcp@latest"]
    }
  }
}